Dan Grigorovici
SemanticWeb.com Contributor

NEW YORK — Recently, I talked to a lot of key VC principals, and they confirmed what I have been suspecting for a while: the Semantic Web (or “Linked Data,” “dataweb,” etc.) has a credibility problem, to the point of being suggested that it’s better to avoid “artificial intelligence,” “semantic” keywords in executive summaries, for fear of hitting a disbelief (“the Semantic Web is the future of the web and it will always be” proudly since 2001).

To comment on this article, go to the Datamation Blog.

Coming from me — a deep believer in a “smarter” web as the future of the Internet and having worked with, and working on a related startup — this is a pretty big deal, but I think we need the community to focus on the business problem and face the issues in order to solve them. I have thus decided to make this the first topic of what will become a series of posts, because I think the SemWeb community needs to face the challenges coming from the business crowd and address them. I am hoping this will spur a serious debate and work on solving them.

I am a “data geek,” a Semantic Web entrepreneur, technologist, and believer. But I don’t think solving the Semantic Web credibility issue and designing its “killer app” and consumer success has anything to do with technology. I argue here that it’s the lack of business focus and basic ability to answer some simple questions (which non-semantic apps can) that are at the core of the continued lack of realization of even the smallest sign of a killer Semantic Web app.

There are some great startups (Twine, Adaptive Blue, Freebase, Zitgist, and others) that rose in the recent years, surely; there are some technological reasons why adoption has been slow, surely. But at its core, I don’t think solving the problem and delivering the vision of a “smarter” web has much to do with the web site owners’ laziness in adopting a standard, or the scalability of triple stores, or lack of technical expertise from the part of consumers, clients, funders, etc.

The point is: After more than seven years of promising the Semantic Web deliverance, we still can’t get our one-pagers clear. We still can’t explain our proposition to users, funders, or anyone else outside the community, what we are building and how it is better than the “dumb” (but increasingly crowd-telligent) and newspaper-ish web of today. That is, we can’t do any of the above without needing to dive into a long exhortation into the meanders of technological detail, or without advocating the need to build vertical knowledge bases from the ground up. At which point we lose both the user, and the funder. Not to mention the political factions that exists in the community today between the practical and the purists (more about this in my next post).

One of the best and most succinct presentations of some of the issues comes from Nick Lothian, commenting on Peter Norvig, author of one of the best AI textbooks and director of research at Google. The other commentary I have come across, much more focused on technology issues, is, “The 7 (f)laws of the Semantic Web.”

Let me explain what the issue is by linking my personal experience and thinking, in detail. I believe the Semantic Web community sounds a lot like a solution in search of a problem. It is only natural to be so, since the work has been done mostly by technologists, but this has been hampering the ability to generate the “killer app.” How so, you will ask, in disbelief? Here are my answers:

“Where’s the (business) beef?”

From the point of view of technology, we are almost there: having arisen from the academic community and to a large extent (except implementations in corporate projects) limited to (still) academic projects, Semantic Web projects don’t lack a host of technological implementation choices. What we do lack is a consistent business team in every one of these projects. Please enter your content here.

Most of the current efforts are still lead by technologically deep teams, without business direction. What I mean is that most of these projects arise from deep evangelical SemWeb passion from developers, without much direction in terms of understanding and connecting with the consumer and business environment, from the basic points of view of product positioning, business strategy, marketing, sales, etc. Nick Lothian mentions how most attempts to get answers from the community end up in a “religious conversion” attempt without really answering the problem, and that is my personal experience too. I have seen people who always address the credibility problem by arguments of the form: “current web is broken, we need to do x and y,” (technologically, mostly). We all know the current web is not what it could/should be, but we tend to hear and dismiss the criticisms to the Semantic Web (see “Semantic Web skeptical reactions”) with more proselytizing, rather than listening to our critics and work on our “pitch.”

What we need is a healthy dose of business perspective on things, and attempt to answer the following: are we trying to solve a new problem, or an old problem? If the latter, is it because our technology is better than the current one, or because we are redefining the problem in terms of our technology in a very convenient and self-serving way? What do we need the disbelievers to do to experience how much better our technology is? How do we get to adoption without asking our disbelievers to completely change their Internet behaviors?

Essential questions

If you notice, most of these questions are no different than the typical questions you might get after delivering any “elevator pitch” (whether semantic or not) to any investor. Interestingly, very little of the public rhetoric of the Semantic Web community really addresses them, and it makes me think: why should the essential Web 3.0 questions we need to answer be different than the Web 2.0, or even worse/better, Web 1.0? I believe they should NOT be different, and the fact that we’re not addressing them is one of the sources of why the Semantic Web has been “talking about trying to solve easy problems since 2001,” (Peter Norvig, from Nick Lothian’s blog post).

All of us trying to build a Semantic startup in hopes of being part of the next killer app wave (including myself), we should make it a must to hire/motivate/associate with and learn something from great business problem solvers, whether sales/marketing, business strategists, product managers, who are purposely non-semantic web evangelists. If we are trying to convince the disbelievers, we should really follow Rhetoric 101: we should not ask our audience to rise up to our (technical) arguments, but we should be the ones adapting our logic to answer the questions posed in theirs.

And the truth is: answering any of these questions has nothing to do with Semantic Web, or even technology in general. Why? Because technology is a ladder helping us in solving a non-technology problem, and not a goal in itself. In other words: the problems the Semantic Web technology is solving are not semantic, nor complex or technological; they are simple: how do I get to the information I need without having to navigate on 100+ pages, how is this better for both consumers and businesses.

We need to be able to answer them without launching in lengthy exhortations of RDF, SPARQL, as supporting arguments. We need to use the same language as our disbelievers do, and having them (want to) come to us because they cannot do it without. Now, that IS what a killer app should be, is it not? My experience is that in our working so hard on the technology (and becoming so focused on insuring it delivers), we have forgotten to build what’s essential: the business and consumer “hooks” — the simple, the necessary.

How do we do it? Here are some thoughts; I invite all my readers to contribute, dismiss (with arguments), oppose, enhance, etc. Only this way we can get our “pitch” to be more effective. Check back tomorrow for part two of this column.

1. Find out the business problem the Semantic Web is solving

The Semantic Web seems to many critics a “heavy” proposition: They are told they need to enrich their data with semantic meta-data/markup, rewrite their existing sites, to open up and break the data silos, etc. These are difficult requirements from a business environment that is really not that open, as most of the competitiveness still lies in ownership of data, standards, etc. (open APIs, Web Services, etc., have only begun to be en vogue, and the richest data is still closed).

This is precisely why the Semantic Web has been successful in the enterprise environment much more than “out there” on the consumer Web. Because businesses still believe that maintaining control of their data is their best asset (as opposed to opening their data, but maintaining control to processing, enriching, etc.). To use an example, we should ask ourselves (when working on our pitch for businesses and/or partners): Why would Walmart open their data to eBay, or any other competitors? Surely, we have built Linked Data from a lot of open source data sets (SIOC, Dbpedia, MusicBrainz, etc.), but the real challenge and success will happen when we have successfully convinced the Walmarts that opening their data is actually beneficial to their business.

Trying to answer this problem has implications on data and user privacy, competitiveness, and it may require us to think less semantically and more like the people we are trying to convince. But isn’t this the best way to win over your “audience?” We also need to be able to answer how is the Semantic Web improving/solving/addressing the monetization problem on the Internet today, e.g. will a semantic web-based application be monetized through page views, using CPM, CPC, CPA models still, or will it disrupt existing business and revenue models (assuming there is a problem with it — I think there is, galore). If we tie the answer to the latter question, to the answer to the first, we got ourselves a winner. These are things we can’t quite answer today outside of closed enterprise semantic implementations or limited academic/research projects.

While working on answers to the questions above, we also need to come up with why are we better than what exists today. Rather than dismissing our “competitors” (which tend to be the entire Web 1.0 and 2.0), we should really study them and be ready with answers to questions that don’t require reading extensively AI literature or convincing site owners to add markup to their pages simply because we are telling them it will be better. We need to start with the business problem, and end with the semantic web solution, not the other way around as we have been doing “proudly, since 2001.”
2. Find out the consumer problem the Semantic Web is solving

Let’s face it: the consumer/user does NOT care about the Semantic Web. They only care about things being made easier for them, while being provided an immediate value to them, ideally without displacing their current behavior too much. As naïve as it sounds (to a developer), “what can Semantic Web do for me” is a valid question that needs a good answer.

This is where conversations about the “Semantic Web” (capitalized for purism and rich with meta-data) and “semantic web” (lower case, more practical, based on statistical and machine learning) are relevant, because as long as either one of the two improves information demands of the user/consumer, it really does not matter from her perspective which avenue is taken. I argue here that, by the same token we have forgotten about the business problem, we also forgot about the consumer. This is the same consumer who can make or break the adoption of our killer app, and needs not be evangelized with teachings of RDF. My current perception of the childish sort of conversation with me as a user from many a Semantic Web application goes a little like this:

SemWeb app: Dan, you really should use our application, because you are in full control, can get your very granular information search in seconds, and can query the web like a database.

Dan: OK, that sounds nice. What do I need to do to use this?

SemWeb app: Well, you need to learn SPARQL, a bit of RDF so you can type proper query syntax for our engine; you might want to also install this (mostly) Firefox widget so that you can visualize your query results nicely. But wait, there’s more: Each site you are querying might use its own schema that may not always map from one another completely so there is the chance you may get some junk as results, etc. etc.

Dan (from both the perspective of a user and SemWeb enthusiast): Well, I think I am fine spending time and lots of pages the way I have been.

My point behind the little drama above is that even though I am well connected to almost all consumer applications being developed recently and having been testing most of them, I did not find one that I used beyond its first trials. There must be a reason why that is: perhaps because we didn’t get our consumer story quite straight yet.

Check back tomorrow for part 2.

Dan Grigorovici is an authority on the semantic web and advertising. He is vice president of data strategies and analytics at Tacoda, a division of AOL that runs behavioral targeting advertising networks. Prior to Tacoda, Dan worked at Digitas as the vice president/associate director of digital analytics. His startup, Disruptive Logic, is building Life Engine, a consumer cross-channel intelligent agent technology that will personalize the Internet.

Semantic Tech & Business Conference Returns to San Francisco

Semantic Tech & Business Conference returns to San Francisco in June! Join us from June 3-7 for complete coverage of Big Data, Linked Data, Extreme Information Management, and Semantic Web. From breakthrough approaches to solving business problems to the big data implications of fast–evolving technologies, SemTechBiz provides you with an unparalleled interactive experience and delivers tangible business value. We're offering a special early rate when you register by February 17. Sign up now!